Shaping the digital brain for Industry 5.0: a review of industrial world models

Annals of Forest Science · Springer Nature · 2026-11-13 · DOI: 10.1007/s10845-026-03014-y

Аннотация

World models describe environmental state evolution and provide a predictive basis for planning, control, and decision-making. Although they have received increasing attention in artificial intelligence, existing reviews mainly address general world models, embodied intelligence, robotics, and autonomous driving, while systematic synthesis for industrial applications remains limited. This review addresses this gap by defining industrial world models as specialized modeling frameworks for industrial objects under engineering requirements. It clarifies the conceptual foundations, development trajectory, and boundaries of world models, and then presents a formal definition of industrial world models. The review examines their core elements, including state-space representation, action-space representation, model architecture, and taxonomy, with particular attention to specialist world models and world foundation models. Using a micro-to-macro industrial object hierarchy as the organizing perspective, it summarizes representative applications across four industrial layers: mechanism, control, execution, and business, focusing on how industrial world models represent, predict, and simulate industrial state evolution under different engineering objectives. The review further analyzes methodological, deployment, and integration challenges, and then discusses Industry 5.0-oriented directions related to human-centricity, sustainability, and resilience. Overall, this review provides a structured reference for understanding industrial world models and their role in scalable, robust, cross-layer industrial intelligence.

Читать у издателя →

Похожие статьи